1. Field of the Invention
The present invention relates to charge-coupled devices or CCDs. It more specifically aims at a two-phase CCD array device in which charge packets of two adjacent lines are transferred according to opposite directions. A preferred application to a CCD image sensor will be described hereafter, it being understood that the present invention may apply to any type of CCD.
2. Discussion of the Related Art
The charge transfer in a CCD is often carried out in four phases, that is, the shifting of a charge packet from one pixel to an adjacent pixel takes four steps corresponding to four periods of a clock driving the transfer.
FIGS. 1A to 1C schematically show a portion of a four-phase charge-coupled image sensor. FIG. 1A is a top view, FIG. 1B is a cross-section view along plane B-B of FIG. 1A, and FIG. 1C is a cross-section view along plane C-C of FIG. 1A.
An N-type doped layer 3 is arranged on a P-type doped silicon substrate 1. Substrate 1 and layer 3 form the photoconversion area of the sensor. The upper portion of the photoconversion area is divided into a plurality of lines 5 separated by insulation rows 7, for example formed of trenches filled with oxide. Columns of insulated electrodes 9, for example, made of polysilicon, equidistant and perpendicular to lines 5 are arranged above layer 3. A thin oxide layer 11 deposited at the surface of layer 3 insulates electrodes 9 from layer 3. Electrodes 9, properly biased, define in each line 5 a succession of potential wells where electric charges can be stored. In the shown example, a pixel is defined in each line by four successive electrodes G1 to G4. The potential well corresponding to such a pixel is created by application of a high voltage, for example, on the order of 5 V, to electrodes G2 and G3, and of a low voltage, lower than the high voltage, for example, on the order of 0 V, to electrodes G1 and G4.
During an image acquisition period, the sensor is illuminated and electrons resulting from the creation, by absorption of a photon, of an electron-hole pair in the photoconversion area are stored in the potential wells which fill up proportionally to the illumination of the corresponding pixel. The illumination light needs to cross electrodes 9 and insulation layer 11. The thickness of the active region of the sensor, essentially formed by substrate 1 and layer 3, is sufficient to absorb the photons, whatever their wavelengths in the wanted spectrum.
After the acquisition period, a transfer period is provided, during which the charges stored in the potential wells are transferred in the direction indicated by arrows 13, in parallel for the plurality of columns and in series for the pixels of a same line 5, towards read and/or storage circuits. The charge shifting is ensured by successive modifications of the voltages applied to the electrodes.
FIG. 2 schematically illustrates a simple four-phase mode of transfer of the charges from one well to an adjacent well by switching, between high and low states, of voltages Φ1, Φ2, Φ3, Φ4 applied to electrodes G1, G2, G3, G4 of each pixel.
At a time t0 corresponding to the end of an image acquisition period, charges, shown by the hatched areas of the drawings, are stored in the potential wells formed by application of a high voltage on electrodes G2 and G3 and of a low voltage on electrodes G1 and G4.
At a time t0+T, T being the period of the clock driving the charge transfer, the voltages applied to electrodes G2 and G4 are switched. Thus, the shifting of the potential wells causes the synchronized shifting of the charge packets to the right. To ease the transfer, electrode G4 will be set to the high voltage before electrode G2 is set to the low voltage.
At a time t0+2T, the voltages applied to electrodes G1 to G3 are switched. At a time t0+3T, the voltages applied to electrodes G2 and G4 are switched. Finally, at a time t0+4T, the voltages applied to electrodes G1 and G3 are switched.
Thus, at the fourth clock period after time t0, the charges stored in a potential well in front of a pixel have been shifted towards a potential well in front of an adjacent pixel of the same line. At the sensor output, the shifted charge packets may be converted into electric voltages by adapted circuits, to form an image signal.
Of course, the transfer period is short as compared with the acquisition period. As an example, the acquisition period is on the order of from 20 to 50 ms and the electrode switching clock frequency may be greater than 2 MHz, which provides a transfer time shorter than 2 ms for a line of 1,000 pixels and a shifting in four phases.
To decrease the transfer period and to simplify the electrode switching circuits, two-phase charge-coupled devices have been provided.
FIG. 3 schematically shows a portion of an example of a two-phase charge-coupled device. FIG. 3 is a cross-section view along the same plane as FIG. 1B, previously described. The structure of the sensor of FIG. 3 resembles that of the four-phase image sensor described in relation with FIGS. 1A to 1C. In the two-phase sensor, as in the four-phase sensor, a pixel is defined, in each line, by four successive electrodes G1 to G4. Layer 3 is divided into alternating columns of two different doping levels, under electrodes 9. In the shown example, the columns of layer 3 under electrodes G1 and G3 have a doping of a first level N1 and the columns of layer 3 under electrodes G2 and G4 have a doping of a second level N2 greater than N1. Electrodes G1 and G2 on the one hand and G3 and G4 on the other hand are interconnected, for example, by metallization levels, not shown.
FIG. 4 schematically illustrates the storage, during an image acquisition period, of electrons photogenerated in potential wells formed by application of voltages Φ1, Φ2 to electrodes G1, G2, G3, G4 of each pixel. FIG. 4 further illustrates a simple two-phase mode of electron transfer, from one well to an adjacent well, by switching between high and low states of voltages Φ1 and Φ2.
At a time t0 corresponding to the end of an image acquisition period, charges shown by the hatched areas of the drawing are stored in potential wells formed by application of a low voltage, for example, on the order of 0 V, on electrodes G1 and G2 and of a high voltage, for example, on the order of 5 V, on electrodes G3 and G4. When two adjacent electrodes are set to a same voltage, the photogenerated electrons are stored in the corresponding N-layer portion of highest doping level (level N2 of FIG. 3). Further, when two adjacent electrodes are set to respectively high and low voltages, the electrons are stored in the N layer region under the high-voltage electrode. Thus, at time t0, charge packets corresponding to points of the acquired image are stored in the N layer, mainly under electrodes G4 of each pixel.
At a time t0+T, T being the period of the clock driving the charge transfer, voltages Φ1 and Φ2 applied to electrodes G1, G2 and G3, G4 are switched to increase the electrostatic potential under electrodes G1 and G2 and to decrease the electrostatic potential under electrodes G3 and G4. This results in a displacement of the potential wells, which causes the synchronized shifting of the charge packets to the right. Thus, at the second clock period after time t0, the charges stored in a potential well under a pixel have been shifted towards a potential well under an adjacent pixel of the same line.
As an example, for an electrode switching clock frequency greater than 2 MHz, the transfer time is shorter than 1 ms for a line of 1,000 pixels and a shifting in two phases.
A disadvantage of two-phase CCDs of the type described in relation with FIG. 3 is that they are more difficult to form than four-phase sensors. Indeed, the adjacent columns of the two-phase sensor differ by their doping level. For the manufacturing of such a sensor, two successive electrode-forming sequences need to be provided. In a first sequence, a first alternation of electrodes G1, G3 (FIG. 3) is formed, at the surface of a uniformly-doped N layer at a first level N1. An implantation step is then carried out to obtain, in the regions of the N layer unmasked by first electrodes G1 and G3, a doping level N2 greater than N1. Then, in a second sequence, a second alternation of electrodes G2, G4, interposed between first electrodes G1 and G3, is formed.
Thus, a disadvantage of four-phase CCD sensors is that the switching mode of the voltages applied to the electrodes is more complex than that of two-phase sensors. A disadvantage of two-phase sensors is that they are more difficult to manufacture than four-phase sensors.
In the two- or four-phase CCD sensors described hereabove, read and/or storage circuits capable of converting the shifted charge packets into electric voltages to form an image signal are provided at the output of the charge transfer lines. Such circuits are bulky. To optimize the positioning of the output circuits and to decrease the general bulk, bidirectional CCDs capable of transferring adjacent pixel lines along opposite directions have been provided. To form such circuits, it is provided to divide each pixel line into two separate lines and to double the number of electrodes. This enables to simultaneously switch separate electrodes according to separate sequences, to control the charge transfer in two adjacent half-lines according to opposite directions.
A disadvantage of such bidirectional sensors is that the multiplication of the number of electrodes makes them more complex to manufacture than unidirectional CCD sensors. Another disadvantage of such sensors is that the switching of the voltages applied to the electrodes is more complex than that of unidirectional sensors.
A general disadvantage of the above-described CCD sensors is that the light needs to cross the polysilicon transfer control electrodes. Part of the photons are thus absorbed in the electrodes, which decreases the sensor sensitivity. To overcome this disadvantage, the transfer electrodes may be arranged next to the photoconversion region rather than above it. However, this solution has the disadvantage of increasing the bulk for a given size of the photoconversion region.
Another general disadvantage of the above-described CCD sensors lies in the fact that the charge storage capacity associated with each pixel is limited by the electrode surface area and by possible carrier recombinations.